EC 703:
Image Processing and
Computer Vision
By:
Dr. Hemant S. Goklani
ECE Department
IIIT, Surat
UNIT 2
Image Enhancement in Spatial Domain
and Frequency Domain
UNIT2_1:
Image Enhancement in the Spatial Domain
Image Enhancement
Goal: process an image so that the result is more
suitable than the original image for a specific
application
Visual interpretation
Problem oriented
Image enhancement example
Two categories
There is no general theory of image
enhancement
Spatial domain
Direct manipulation of pixels
Point processing
Neighborhood processing
Frequency domain
Modify the Fourier transform of an image
Outline: spatial domain operations
Background
Gray level transformations
Arithmetic/logic operations
Background
Spatial domain processing
g(x,y)=T[ f(x,y) ] f(x,y) g(x,y)
T
f(x,y): input image
g(x,y): output image
T: operator
Defined over some neighborhood of (x,y)
Background (cont.)
* T applies to each pixel
in the input image
* T operates over
neighborhood of (x,y)
Point processing
1x1 neighborhood
Gray level transformation, or point processing
s = T(r) Where r is values of pixel before process
s is values of pixel after process
T is a transformation that maps a pixel value
r into a pixel value s.
contrast
stretching thresholding
Neighborhood processing
A larger predefined neighborhood
Ex. 3x3 neighborhood
mask, filters, kernels, templates, windows
Mask processing or filtering
Some Basic Gray Level Transformations
Image negatives (complement)
Log transformation
Power-law transform
Piece-wise linear transform
Gray level slicing
Bit plane slicing
Some gray level transformations
•Lookup table
•Functional form
Image Negatives
The negative of an image with gray L levels is
given by the expression
s=L–1–r
Where r is value of input pixel, and
s is value of processed pixel
Suitable for images with dominant black areas
Log Transformations
The general form of the log transformation shown
in
s = c log(1+r)
Where c is constant, and it is assumed that r 0
Compress the dynamic range of images with large variation
in pixel values
C = 1.0
C = 0.8
Example: Log transformations
log(fft2(I)) : log of Fourier transform
log
2d Fourier transform
Exponential Transformations
The general form of the log transformation
shown in
s = c exp(r)
Where c is constant, and it is assumed that r 0
C = 1.0
C = 0.8
Power-Law Transformations
Power-law transformations have the basic form
s cr
Where c and are positive constant.
Sometime above Equation is written as
s c(r )
= 0.5
= 1.0
= 5.0
>1
<1
: gamma display, printers, scanners follow power-law
Gamma correction
Example: Gamma correction
CRT: intensity-to- =2.5
voltage response
follow a power-law.
1.8<<2.5
=1/2.5
=2.5
Power-law: <1
Expand dark gray
levels
=0.6
=0.4 =0.3
Power-law: >1
Expand light
gray levels
=3
=4 =5
Piece-wise linear transformations
control point
Advantage: the
piecewise function
can be arbitrarily
complex
Contrast Stretching
Low contrast image can result from poor illumination,
lack of dynamic range in the imaging sensor, or even
wrong setting of a lens aperture during image
acquisition.
The idea behind the contrast stretching is to increase
the dynamic range of the gray levels in the image being
processed.
Contrast Stretching
L-1 (r2,s2) The location points (r1,s1)
Output and (r2,s2) control the shape
Gray level of the transformation function
s
T(r)
0 (r1,s1) If r1=s1 and r2=s2, the
0 L-1 transformation is a linear
Input gray level, r function that produces no
changes in gray levels
If r1=r2, s1=0 and s2=L-1 the transformation becomes
thresholding function that creates a binary image
Intermediate values of (r1,s1) and (r2,s2) produces
various degrees of spread in the gray levels of the output image,
Gray level slicing
Output L-1 L-1
Gray level Output D T(r)
T(r)
s D Gray level
s
C
0 0
0 A B L-1 0 A B L-1
Input gray level, r Input gray level, r
• Highlighting a specific range of gray levels in an image often is
desired
•Applications include enhancing features such as masses of
water in satellite imagery and enhancing flows in X-rays images.
Contrast
stretching
Contrast Streching
Gray-level slicing
Highlighting a specific range of gray levels
Bit-plane slicing
Instead of highlighting gray level ranges, highlighting
the contribution made to total image appearance by
specific bits might be desired.
Suppose that each pixel in the image is represented by
8-bits. Imagine that the image is composed off eight 1-
bit planes, ranging from bit-plane 0 for the least
significant bit to bit plane 7 for most significant bit
Bit-plane slicing
* Highlight specific bits
1
0
0
1
0
1
0
0
Ex. 15010 bit-planes of an image
(gray level 0~255)
Bit-plane slicing: example
For image
compression
7 6
5 4 3
2 1 0
Histogram Processing
The histogram of a digital image with gray levels in
the range [0,L-1] is a discrete function
h(rk ) nk
Where, rk is the kth gray level and
nk is the number of pixels in the image having gray level rk
In an image processing context, the histogram of an image
normally refers to a histogram of the pixel intensity values
This histogram is a graph showing the number of pixels in an
image at each different intensity value found in that image
HISTOGRAM
pixels
130
36 36
22
0 1 2 3 level
Image 16 x 14 = 224 pixels
Histogram
The (intensity or brightness) histogram shows how many
times a particular grey level (intensity) appears in an image.
For example, 0 - black, 255 – white
7
0 1 1 2 4 6
5
2 1 0 0 2 4
3
5 2 0 0 4 2
1
1 1 2 4 1 0
0 1 2 3 4 5 6
image histogram
Image Histograms
x-axis – values of intensities
y-axis – their frequencies
Histogram Processing
Histogram h(rk ) nk
rk is the k th intensity value
nk is the number of pixels in the image with intensity rk
nk
Normalized histogram p( rk )
MN
nk : the number of pixels in the image of
size M N with intensity rk
Histogram Processing
Histogram Equalization
Histogram Matching(Specification)
Local Enhancement
Fundamental of random variable
PDF (Probability Density Function) is the
probability of each element
CDF (Cumulative Distribution Function) is
summation of the probability of the element
that value less than or equal this element
PDF
The PDF (probability density function) is denoted
by p(x)
pdf
1/6
1 2 3 4 5 6
CDF
The CDF (cumulative density function) is
denoted by P(x)
cdf
1
1/6
1 2 3 4 5 6
Relation between PDF and CDF
PDF can find from this equation
d [ P( x)]
p ( x) P( x)
dx
CDF can find from this equation
xn
P ( x xn ) p( x)dx
x x0
Histogram Equalization
We focus attention on transformations of the
form
s T (r ) where 0 r 1
That produce a level s for every pixel value r in the
original image.
And we assume that the transformation function T(r)
satisfies the following conditions:
(a) T(r) is single-valued and monotonically increasing
in the interval 0 ≤ r ≤ L - 1; and
(b) 0 ≤ T(r) ≤ L - 1 for 0 ≤ r ≤ L - 1
Reason of Condition
The requirement in (a) that T(r) be single valued is
needed to guarantee that the inverse transformation
will exist, and the monotonicity condition preserves the
increasing order from black to white in the output
image
Condition (b) guarantees that the output gray levels will
be in the same range as the input levels
Transformation function
Level s
sk=T(rk) T(r)
Level r
0 rk 1
Note
The inverse transformation from s back to r is denoted
1
s T (r ) where 0 r 1
There are some cases that even if T(r) satisfies conditions
(a) and (b), it is possible that the corresponding inverse
T-1(s) may fail to be single valued.
Idea of Histogram Equalization
The gray levels in an image may be viewed as random
variables in the interval [0,1]
Let pr(r) denote the pdf of random variable r and
ps(s) denote the pdf of random variable s; if pr(r) and T(r) are
known and T-1(s) satisfies condition (a), the formula should be
dr
p s ( s ) pr ( r )
ds
r
s T (r ) pr ( w)dw
0
Idea of Histogram Equalization
By Leibniz’s rule that the derivative of a definite
integral with respect to its upper limit is simply the
integrand evaluated at that limit
ds dT (r )
dr dr
d
r
pr ( w)dw
dr 0
pr ( r )
Idea of Histogram Equalization
Substituting into the first equation
dr
p s ( s ) pr ( r )
ds
1
pr ( r )
pr ( r )
1, 0 s 1
Idea of Histogram Equalization
The probability of occurrence of gray level rk in an
image is approximated by
nk
pk (rk ) k 0,1,2,..., L 1
n
k
sk T (rk ) pr (rj )
j 0
k nj
k 0,1,2,..., L 1
j 0 n
Histogram Equalization
THANK YOU
